Applicability of controlled natural languages for architecture analysis and documentation: an industrial case study

Authors Sandra Schröder
Georg Buchgeher
Editors L. Duchien
C. Trubiani
R. Scandariato
et al.
Title Applicability of controlled natural languages for architecture analysis and documentation: an industrial case study
Booktitle Proceedings of the 13th European Conference on Software Architecture - Volume 2 (ECSA '19)
Type in proceedings
Publisher ACM
ISBN 978-1-4503-7142-1
DOI 10.1145/3344948.3344981
Month September
Year 2019
Pages 190-196
SCCH ID# 19071
Abstract

Formal approaches are a prerequisite for automated quality control. However, such approaches are often difficult to learn and apply and thus have not found their way into mainstream software development. Controlled natural languages (CNLs) seek to overcome the drawbacks of formal approaches by hiding their formality behind a restricted set of natural language, which is intended to be easier to use. In this paper, we analyze the applicability of CNLs for automated architecture analysis and as a means for architecture documentation in an industrial case study. We evaluated the CNL with 12 experienced practitioners in a focus group. The results show that practitioners perceive CNLs as an appropriate and an understandable means for formalizing and documenting architectural rules. Even without verifying rules automatically, a project can benefit from such languages, since they allow to find a common sense about architecture concepts and relations and to use those concepts and relations consistently throughout the development and evolution phase.